Individuals with hearing loss typically experience great difficulty understanding speech in noisy environments. This is particularly true for an increasing number of elderly people, who often have difficulty carrying on a normal conversation in social situations, such as parties, meetings, sporting events or the like, involving a high level of background noise. Such hearing loss in noise is generally due to reduced hearing sensitivity of the ear, which results in an attenuation of all sounds and a distortion of sounds. In other words, reduced hearing sensitivity causes a listener to perceive speech to be not only softer, but also garbled.
Hearing aids are known and have been developed to assist individuals with hearing loss. Hearing aids generally amplify sounds, and thus compensate for the attenuation effect of reduced hearing sensitivity. However, it is the distortion effect, i.e., the inability of a listener to discriminate between sounds, that makes speech intelligibility in noise difficult, or even impossible, for most people. A solution to improve speech intelligibility in noise, therefore, must compensate for the distortion effect by attenuating background noise in relation to desired speech signals. In fact, several investigations on speech intelligibility in noise have demonstrated that every 4-5 dB attenuation of background noise may raise speech intelligibility by about 50%.
Directional microphones have been used in hearing aids to attenuate background noise. A suitable measure of the directional effect of such a microphone is the directivity index. The directivity index indicates in decibels the amount in which a directional microphone attenuates sounds in a diffuse sound field as compared to an omnidirectional microphone. In the frequency range most important for speech discrimination (i.e., about 500 to 5000 Hz), the directivity index for a typical 1st order directional microphone is only approximately 5 dB. This level of directivity at such frequencies, while an improvement for individuals with mild to moderate hearing loss is insufficient for situations involving more severe loss.
As a result, the use of several directional microphones was proposed by an inventor in the present application for improving the directivity of a hearing aid and thus speech intelligibility under such conditions of noise and hearing loss. See Soede, Willem, “Improvement of Speech Intelligibility in Noise: Development and Evaluation of a New Directional Hearing Instrument Based on Array Technology” Ph.D. Thesis, Delft University of Technology, Delft, The Netherlands, 1990, which is incorporated herein by reference in its entirety (referenced hereinafter as “the Delft Thesis”). The Delft Thesis proposed that traditional microphone array techniques already of use in other fields, such as astronomy, sonar, radar and seismology, could be used in hearing aid applications to improve directional characteristics.
One such traditional microphone array is shown in FIG. 1a. A microphone array system 2 includes a plurality of microphones 4 aligned along an axis x. Each microphone 4 generates a signal from a desired sound typically impinging along an axis y, as well as from undesired sounds from all directions, and each signal generated is transmitted to a processor 6. Each processed signal is then added to produce an amplified output signal 10. Such a microphone array is generally referred to as a “broadside” array.
Another such traditional microphone array is shown in FIG. 1b. A microphone array system 1 includes a plurality of microphones 3 aligned with equal spacing along an axis z. Sound impinges on each of the microphones 3 along the z-axis as shown by arrow 5. Each signal generated by the respective microphones is then transmitted to a processing block 7. Depending on the location of a particular microphone 3 along the z-axis, the processing block 7 may apply a delay to the received signal.
More specifically, for the first microphone, the signal (labeled m4) is delayed four delay periods 8; for the second (labeled m3), the signal is delayed three delay periods 8; for the third (labeled m2), the signal is delayed two delay periods 8; for the fourth (labeled m1), the signal is delayed one delay period 8; and for the last microphone (labeled m0), the signal is not delayed at all. Applying a delay as such ensures that the signals received along an axis z are in phase and thus in condition for maximum summation. Once the signals are in phase, each signal is processed using a processor 9, and then all signals are summed to produce an output signal 11.
For sound impinging on the microphone array system 1 at an angle shown by arrow 5, the total delay period (τm) for any given processing block 7 in FIG. 1b can be calculated using the following formula:
            τ      m        =                  m        ⁢                                  ⁢        Δ        ⁢                                  ⁢        z            c        ,      m    =    0    ,  1  ,  2  ,      3    ⁢    …  where m represents the microphone 3 number, Δz the distance between the microphones 3, and c the velocity of sound. Each equal delay period 8 can thus be calculated as τm/m or Δz/c.
In addition, the number of delay periods 8 for any given number of microphones in an array can be calculated using the following formula:
      n    ⁡          (              n        -        1            )        2where n is the number of microphones in the array. Thus, for the array in FIG. 1 having five microphones, 10 delay periods 8 are required.
The operation of the microphone array system 1 of FIG. 1b can be demonstrated graphically as shown in FIG. 2. Each microphone 3 generates a signal 13 from impinging sound energy at a particular time along an axis t. Each generated signal 13, except for the last one corresponding to microphone m0, is delayed a period tau (τ) as discussed above so that the delayed signals 13 are in phase. The signals 13 are then added to produce an output signal 15 (corresponding to signal 11 in FIG. 1). Because the same delay period τ is applied to all sound received from a direction to the rear of the microphone array, the resulting summed signal received from rear sounds is out of phase (see signal 17). What results, therefore, is an amplified signal with a preference for all sounds coming from the front of the array. In other words, the array achieves a much higher directivity than is possible with the use of only a single microphone 3. An array of microphones of the type discussed above with respect to FIGS. 1b and 2 above is generally referred to as an “endfire” array.
In the 1930's, Hansen and Woodyard, working with large arrays (approximately 10 times the wavelength) in radar applications, derived a formula mathematically for optimizing the directivity of such endfire arrays. The Delft Thesis mentioned above applied the Hansen and Woodyard principle to acoustics and determined that the time delay τ set forth above can be optimized using the following formula:
            τ      m        =                            m          ⁢                                          ⁢          Δ          ⁢                                          ⁢          z                c            ⁢              (                  1          +          ɛ                )              ,            with      ⁢                          ⁢      ɛ        =                            2.94          ⁢          λ                          2          ⁢          π          ⁢                                          ⁢          L                    =                        2.94          ⁢          c                          2          ⁢          π          ⁢                                          ⁢          fL                    where λ equals the sound wavelength, L equals the array length, and f equals the sound frequency. This mathematical Hansen-Woodyard optimization for endfire arrays set forth in the Delft Thesis is plotted in FIG. 3 (as directivity index versus frequency—see curve 19). The traditional approach (i.e., prior to optimization) is also shown in FIG. 3 as curve 18. The mathematical Hansen-Woodyard optimization is further plotted in FIG. 4 (as delay time versus frequency—see curve 21). The traditional approach (i.e., prior to optimization) is also shown in FIG. 4 as curve 20.
In implementing a microphone array system to match the mathematical Hansen-Woodyard optimization, however, the Delft Thesis fell short of achieving such optimization (see curves 23 and 25 in FIGS. 3 and 4, respectively). Over a frequency range of approximately 250 to 6000 Hz, the Delft Thesis produced an average directivity index of approximately 8.1 dB. While this was an improvement over the traditional approach (which yielded an articulation index weighted directivity index (“AIDI”) of approximately 6.7 dB), the Delft Thesis simply did not match the AIDI of 10.2 dB achieved by the mathematical Hansen-Woodyard optimization.
Thus, it is an object of the present invention to provide a microphone array system that more clearly matches the mathematical Hansen-Woodyard optimization.
It is a further object of the present invention to provide a miniature microphone array system more suitable than traditional arrays for hearing aid applications.
It is yet a further object of the invention to provide a new directivity optimization for short endfire microphone arrays.